Measurement of borneol based on infrared spectrometry

The infrared spectrometry contains multiple information of the sample, and it is easy to be applied to online measurement. To Chinese medicine, this technology can improve the standard of quality control and accelerate the modernization course. In this paper, we investigate the spectral characteristics of borneol, an effective ingredient in many Chinese medicines. The following results are achieved. In middle infrared (MIR) region, utilizing the linear relationship between absorption and concentration, the concentration of borneol with relative error within 4.30% in the strongest absorption region (2950 - 2970 cm-1) is measured; in near infrared (NIR) region, the predicted concentrations of borneol are calculated by using partial least squares (PLS) regression analysis, in which the wavelengths are selected by genetic algorithm (GA) from the absorption bands of borneol in NIR region. The predicted relative error of calibration model is less than 2%. This result shows that PLS regression analysis combinin     

基于二维相关近红外谱参数化及BP神经网络的掺杂牛奶鉴别

将二维相关近红外谱参数化方法与BP神经网络结合,建立掺杂牛奶与纯牛奶的判别模型。分别配制含有尿素牛奶(1～20 g·L-1)和三聚氰胺牛奶(0.01～3 g·L-1)样品各40个。研究了纯牛奶、掺杂牛奶的二维相关近红外谱特性,在此基础上,分别提取了各样品二维相关同步谱的5个特征参数。将这5个特征参数作为BP神经网络的输入,分别建立掺杂尿素、掺杂三聚氰胺、两种掺杂牛奶与纯牛奶的判别模型,采用这些模型对未知样品进行预测,其预测正确率分别为95%,100%和96.7%。研究结果表明：该方法有效地提取了牛奶中掺杂目标物的特征光谱信息,同时又减少了BP神经网络输入变量的维数,实现了掺杂牛奶与纯牛奶的鉴别。